Michael Paquier <michael(dot)paquier(at)gmail(dot)com> writes:
> This is obviously incorrect because recovery_min_apply_delay has been only
> introduced in 9.4. The culprit is visibly the commit message of 8049839 and
> others that mentioned the parameter, though the patch applied does nothing
> about it. Please see attached a patch to fix that.
Good catch! That's on me I guess for not checking what the patch had done
in the back branches.
I didn't like simply deleting any description of the patch's effects, though.
Instead I made it talk about recovery_target_xid, which does exist in all
those branches.
regards, tom lane